PICDGI: A framework for predicting cancer driver genes through dynamic gene-gene interaction modeling of single-cell...

Area of Science:

• Genomics
• Computational Biology
• Cancer Research

Background:

• Identifying cancer driver genes (CDGs) is crucial for understanding tumor progression.
• Traditional frequency-based mutation analysis often overlooks rare but functionally significant regulators.
• Dynamic gene interactions and regulatory roles are key to uncovering novel CDGs.

Purpose of the Study:

• To introduce PICDGI, a novel computational framework for predicting driver-like regulatory genes.
• To leverage dynamic gene-gene interaction modeling with single-cell RNA sequencing (scRNA-seq) data.
• To identify rare and context-specific regulatory drivers of cancer progression.

Main Methods:

• PICDGI integrates dynamic gene-gene interaction modeling with scRNA-seq data.
• It employs a time-varying state-space model with variational Bayesian inference and MCMC sampling.
• Infers functional driver activity from time-resolved expression patterns and regulatory influence.

Main Results:

• PICDGI successfully identified known oncogenes and tumor suppressors in lung adenocarcinoma (LUAD).
• Novel candidate drivers like JPH1 and CHEK1, involved in key cellular processes, were nominated.
• The framework demonstrated superior performance compared to spatial autocorrelation methods and validated on pediatric AML data.

Conclusions:

• PICDGI offers a generalizable and biologically grounded approach for identifying rare cancer drivers.
• The framework enhances understanding of tumor progression by integrating dynamic regulatory insights.
• PICDGI has broad applicability across diverse cancer types for discovering critical regulatory genes.